feat: add conditions to deployment chain configuration

[mugiwaraluffy56]

What would you like to be added:

a way to configure conditions on deployment chain triggers so that downstream apps only get deployed based on the outcome of the upstream deployment. right now a chain fires all matchers unconditionally regardless of whether the first deployment succeeded or failed.

there's actually already a commented-out stub for this in the codebase at pkg/configv1/application.go:

// TODO: Add conditions to deployment chain configuration.
// Conditions *DeploymentChainTriggerCondition `json:"conditions,omitempty"`

something like this would make sense:

pipeline:
  stages:
    - name: K8S_SYNC
  chain:
    conditions:
      status: SUCCESS
    applicationMatchers:
      - name: myapp-prod

Why is this needed:

without this, a failed staging deployment still triggers production. the whole point of chaining deployments is to gate on the previous result, and right now that gate doesn't exist. you'd have to manually cancel downstream deployments after a failure which defeats the purpose.

[mugiwaraluffy56]

@riahtgl can i work on this :)

[khanhtc1202]

I thought I implemented this like that already, even if we have these conditions configurations, what is the point of configuring to trigger the next deployments in case the current deployment is not successful (aka. conditions.status != success) 🤔

As the docs mentioned the behavior of the pipedv0 is as follows
ref: https://pipecd.dev/docs-v0.55.x/user-guide/managing-application/deployment-chain/#:~:text=Once%20a%20node%20in%20a%20block%20has%20finished

[khanhtc1202]

Reading deeper into the config file (ref: https://github.com/pipe-cd/pipecd/blob/master/pkg/configv1/application.go#L454-L456), I now remember the context.
The idea is to handle case like "you want to ONLY trigger the post sync chain in case the first deployment fits some conditions - like having a commit prefix", not like I planned to handle "conditions.status" to ensure only trigger the post sync chain on successful, since (as I mentioned in the comment above) there is no mean to trigger the chain if any of the deployment in chain return failure. WDYT?